email_in.pl

#!/usr/bin/perl -w
# -*- Mode: perl; indent-tabs-mode: nil -*-
#
# The contents of this file are subject to the Mozilla Public
# License Version 1.1 (the "License"); you may not use this file
# except in compliance with the License. You may obtain a copy of
# the License at http://www.mozilla.org/MPL/
#
# Software distributed under the License is distributed on an "AS
# IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or
# implied. See the License for the specific language governing
# rights and limitations under the License.
#
# The Original Code is the Bugzilla Inbound Email System.
#
# The Initial Developer of the Original Code is Akamai Technologies, Inc.
# Portions created by Akamai are Copyright (C) 2006 Akamai Technologies, 
# Inc. All Rights Reserved.
#
# Contributor(s): Max Kanat-Alexander <mkanat@bugzilla.org>

use strict;
use warnings;

# MTAs may call this script from any directory, but it should always
# run from this one so that it can find its modules.
BEGIN {
    require File::Basename;
    chdir(File::Basename::dirname($0)); 
}

use lib qw(. lib);

use Data::Dumper;
use Email::Address;
use Email::Reply qw(reply);
use Email::MIME;
use Email::MIME::Attachment::Stripper;
use Getopt::Long qw(:config bundling);
use Pod::Usage;
use Encode;

use Bugzilla;
use Bugzilla::Bug;
use Bugzilla::Constants qw(USAGE_MODE_EMAIL);
use Bugzilla::Error;
use Bugzilla::Mailer;
use Bugzilla::User;
use Bugzilla::Util;

#############
# Constants #
#############

# This is the USENET standard line for beginning a signature block
# in a message. RFC-compliant mailers use this.
use constant SIGNATURE_DELIMITER => '-- ';

# $input_email is a global so that it can be used in die_handler.
our ($input_email, %switch);

####################
# Main Subroutines #
####################

sub parse_mail {
    my ($mail_text) = @_;
    debug_print('Parsing Email');
    $input_email = Email::MIME->new($mail_text);
    
    my %fields;

    # Email::Address->parse returns an array
    my ($reporter) = Email::Address->parse($input_email->header('From'));
    $fields{'reporter'} = $reporter->address;
    my $summary = $input_email->header('Subject');
    if ($summary =~ /\[Bug (\d+)\](.*)/i) {
        $fields{'bug_id'} = $1;
        $summary = trim($2);
    }

    my ($body, $attachments) = get_body_and_attachments($input_email);
    if (@$attachments) {
        $fields{'attachments'} = $attachments;
    }

    debug_print("Body:\n" . $body, 3);

    $body = remove_leading_blank_lines($body);
    my @body_lines = split("\n", $body);

    # If there are fields specified.
    if ($body =~ /^\s*@/s) {
        my $current_field;
        while (my $line = shift @body_lines) {
            # If the sig is starting, we want to keep this in the 
            # @body_lines so that we don't keep the sig as part of the 
            # comment down below.
            if ($line eq SIGNATURE_DELIMITER) {
                unshift(@body_lines, $line);
                last;
            }
            # Otherwise, we stop parsing fields on the first blank line.
            $line = trim($line);
            last if !$line;
            
            if ($line =~ /^@(\S+)\s*=\s*(.*)\s*/) {
                $current_field = lc($1);
                # It's illegal to pass the reporter field as you could
                # override the "From:" field of the message and bypass
                # authentication checks, such as PGP.
                if ($current_field eq 'reporter') {
                    # We reset the $current_field variable to something
                    # post_bug and process_bug will ignore, in case the
                    # attacker splits the reporter field on several lines.
                    $current_field = 'illegal_field';
                    next;
                }
                $fields{$current_field} = $2;
            }
            else {
                $fields{$current_field} .= " $line";
            }
        }
    }


    # The summary line only affects us if we're doing a post_bug.
    # We have to check it down here because there might have been
    # a bug_id specified in the body of the email.
    if (!$fields{'bug_id'} && !$fields{'short_desc'}) {
        $fields{'short_desc'} = $summary;
    }

    my $comment = '';
    # Get the description, except the signature.
    foreach my $line (@body_lines) {
        last if $line eq SIGNATURE_DELIMITER;
        $comment .= "$line\n";
    }
    $fields{'comment'} = $comment;

    debug_print("Parsed Fields:\n" . Dumper(\%fields), 2);

    return \%fields;
}

sub post_bug {
    my ($fields_in) = @_;
    my %fields = %$fields_in;

    debug_print('Posting a new bug...');

    my $cgi = Bugzilla->cgi;
    foreach my $field (keys %fields) {
        $cgi->param(-name => $field, -value => $fields{$field});
    }

    $cgi->param(-name => 'inbound_email', -value => 1);

    require 'post_bug.cgi';
}

sub process_bug {
    my ($fields_in) = @_; 

    my %fields = %$fields_in;

    my $bug_id = $fields{'bug_id'};
    $fields{'id'} = $bug_id;
    delete $fields{'bug_id'};

    debug_print("Updating Bug $fields{id}...");

    my $bug = Bugzilla::Bug->check($bug_id);

    if ($fields{'bug_status'}) {
        $fields{'knob'} = $fields{'bug_status'};
    }
    # If no status is given, then we only want to change the resolution.
    elsif ($fields{'resolution'}) {
        $fields{'knob'} = 'change_resolution';
        $fields{'resolution_knob_change_resolution'} = $fields{'resolution'};
    }
    if ($fields{'dup_id'}) {
        $fields{'knob'} = 'duplicate';
    }

    # Move @cc to @newcc as @cc is used by process_bug.cgi to remove
    # users from the CC list when @removecc is set.
    $fields{'newcc'} = delete $fields{'cc'} if $fields{'cc'};

    # Make it possible to remove CCs.
    if ($fields{'removecc'}) {
        $fields{'cc'} = [split(',', $fields{'removecc'})];
        $fields{'removecc'} = 1;
    }

    my $cgi = Bugzilla->cgi;
    foreach my $field (keys %fields) {
        $cgi->param(-name => $field, -value => $fields{$field});
    }
    $cgi->param('longdesclength', scalar $bug->longdescs);

    require 'process_bug.cgi';
}

######################
# Helper Subroutines #
######################

sub debug_print {
    my ($str, $level) = @_;
    $level ||= 1;
    print STDERR "$str\n" if $level <= $switch{'verbose'};
}

sub get_body_and_attachments {
    my ($email) = @_;

    my $ct = $email->content_type || 'text/plain';
    debug_print("Splitting Body and Attachments [Type: $ct]...");

    my $body;
    my $attachments = [];
    if ($ct =~ /^multipart\/alternative/i) {
        $body = get_text_alternative($email);
    }
    else {
        my $stripper = new Email::MIME::Attachment::Stripper(
            $email, force_filename => 1);
        my $message = $stripper->message;
        $body = get_text_alternative($message);
        $attachments = [$stripper->attachments];
    }

    return ($body, $attachments);
}

sub get_text_alternative {
    my ($email) = @_;

    my @parts = $email->parts;
    my $body;
    foreach my $part (@parts) {
        my $ct = $part->content_type || 'text/plain';
        my $charset = 'iso-8859-1';
        # The charset may be quoted.
        if ($ct =~ /charset="?([^;"]+)/) {
            $charset= $1;
        }
        debug_print("Part Content-Type: $ct", 2);
        debug_print("Part Character Encoding: $charset", 2);
        if (!$ct || $ct =~ /^text\/plain/i) {
            $body = $part->body;
            if (Bugzilla->params->{'utf8'} && !utf8::is_utf8($body)) {
                $body = Encode::decode($charset, $body);
            }
            last;
        }
    }

    if (!defined $body) {
        # Note that this only happens if the email does not contain any
        # text/plain parts. If the email has an empty text/plain part,
        # you're fine, and this message does NOT get thrown.
        ThrowUserError('email_no_text_plain');
    }

    return $body;
}

sub remove_leading_blank_lines {
    my ($text) = @_;
    $text =~ s/^(\s*\n)+//s;
    return $text;
}

sub html_strip {
    my ($var) = @_;
    # Trivial HTML tag remover (this is just for error messages, really.)
    $var =~ s/<[^>]*>//g;
    # And this basically reverses the Template-Toolkit html filter.
    $var =~ s/\&amp;/\&/g;
    $var =~ s/\&lt;/</g;
    $var =~ s/\&gt;/>/g;
    $var =~ s/\&quot;/\"/g;
    $var =~ s/&#64;/@/g;
    # Also remove undesired newlines and consecutive spaces.
    $var =~ s/[\n\s]+/ /gms;
    return $var;
}


sub die_handler {
    my ($msg) = @_;

    # In Template-Toolkit, [% RETURN %] is implemented as a call to "die".
    # But of course, we really don't want to actually *die* just because
    # the user-error or code-error template ended. So we don't really die.
    return if $msg->isa('Template::Exception') && $msg->type eq 'return';

    # If this is inside an eval, then we should just act like...we're
    # in an eval (instead of printing the error and exiting).
    die(@_) if $^S;

    # We can't depend on the MTA to send an error message, so we have
    # to generate one properly.
    if ($input_email) {
       $msg =~ s/at .+ line.*$//ms;
       $msg =~ s/^Compilation failed in require.+$//ms;
       $msg = html_strip($msg);
       my $from = Bugzilla->params->{'mailfrom'};
       my $reply = reply(to => $input_email, from => $from, top_post => 1, 
                         body => "$msg\n");
       MessageToMTA($reply->as_string);
    }
    print STDERR "$msg\n";
    # We exit with a successful value, because we don't want the MTA
    # to *also* send a failure notice.
    exit;
}

###############
# Main Script #
###############

$SIG{__DIE__} = \&die_handler;

GetOptions(\%switch, 'help|h', 'verbose|v+');
$switch{'verbose'} ||= 0;

# Print the help message if that switch was selected.
pod2usage({-verbose => 0, -exitval => 1}) if $switch{'help'};

Bugzilla->usage_mode(USAGE_MODE_EMAIL);


my @mail_lines = <STDIN>;
my $mail_text = join("", @mail_lines);
my $mail_fields = parse_mail($mail_text);

my $username = $mail_fields->{'reporter'};
# If emailsuffix is in use, we have to remove it from the email address.
if (my $suffix = Bugzilla->params->{'emailsuffix'}) {
    $username =~ s/\Q$suffix\E$//i;
}

my $user = Bugzilla::User->new({ name => $username })
    || ThrowUserError('invalid_username', { name => $username });

Bugzilla->set_user($user);

if ($mail_fields->{'bug_id'}) {
    process_bug($mail_fields);
}
else {
    post_bug($mail_fields);
}

__END__

=head1 NAME

email_in.pl - The Bugzilla Inbound Email Interface

=head1 SYNOPSIS

 ./email_in.pl [-vvv] < email.txt

 Reads an email on STDIN (the standard input).

  Options:
    --verbose (-v) - Make the script print more to STDERR.
                     Specify multiple times to print even more.

=head1 DESCRIPTION

This script processes inbound email and creates a bug, or appends data
to an existing bug.

=head2 Creating a New Bug

The script expects to read an email with the following format:

 From: account@domain.com
 Subject: Bug Summary

 @product = ProductName
 @component = ComponentName
 @version = 1.0

 This is a bug description. It will be entered into the bug exactly as
 written here.

 It can be multiple paragraphs.

 -- 
 This is a signature line, and will be removed automatically, It will not
 be included in the bug description.

The C<@> labels can be any valid field name in Bugzilla that can be
set on C<enter_bug.cgi>. For the list of required field names, see 
L<Bugzilla::WebService::Bug/Create>. Note, that there is some difference
in the names of the required input fields between web and email interfaces, 
as listed below:

=over

=item *

C<platform> in web is C<@rep_platform> in email

=item *

C<severity> in web is C<@bug_severity> in email

=back

For the list of all field names, see the C<fielddefs> table in the database. 

The values for the fields can be split across multiple lines, but
note that a newline will be parsed as a single space, for the value.
So, for example:

 @short_desc = This is a very long
 description

Will be parsed as "This is a very long description".

If you specify C<@short_desc>, it will override the summary you specify
in the Subject header.

C<account@domain.com> must be a valid Bugzilla account.

Note that signatures must start with '-- ', the standard signature
border.

=head2 Modifying an Existing Bug

Bugzilla determines what bug you want to modify in one of two ways:

=over

=item *

Your subject starts with [Bug 123456] -- then it modifies bug 123456.

=item *

You include C<@bug_id = 123456> in the first lines of the email.

=back

If you do both, C<@bug_id> takes precedence.

You send your email in the same format as for creating a bug, except
that you only specify the fields you want to change. If the very
first non-blank line of the email doesn't begin with C<@>, then it
will be assumed that you are only adding a comment to the bug.

Note that when updating a bug, the C<Subject> header is ignored,
except for getting the bug ID. If you want to change the bug's summary,
you have to specify C<@short_desc> as one of the fields to change.

Please remember not to include any extra text in your emails, as that
text will also be added as a comment. This includes any text that your
email client automatically quoted and included, if this is a reply to
another email.

=head3 Adding/Removing CCs

To add CCs, you can specify them in a comma-separated list in C<@cc>.
For backward compatibility, C<@newcc> can also be used. If both are
present, C<@cc> takes precedence.

To remove CCs, specify them as a comma-separated list in C<@removecc>.

=head2 Errors

If your request cannot be completed for any reason, Bugzilla will
send an email back to you. If your request succeeds, Bugzilla will
not send you anything.

If any part of your request fails, all of it will fail. No partial
changes will happen.

There is no attachment support yet.

=head1 CAUTION

The script does not do any validation that the user is who they say
they are. That is, it accepts I<any> 'From' address, as long as it's
a valid Bugzilla account. So make sure that your MTA validates that
the message is actually coming from who it says it's coming from,
and only allow access to the inbound email system from people you trust.

=head1 LIMITATIONS

Note that the email interface has the same limitations as the
normal Bugzilla interface. So, for example, you cannot reassign
a bug and change its status at the same time.

The email interface only accepts emails that are correctly formatted
perl RFC2822. If you send it an incorrectly formatted message, it
may behave in an unpredictable fashion.

You cannot send an HTML mail along with attachments. If you do, Bugzilla
will reject your email, saying that it doesn't contain any text. This
is a bug in L<Email::MIME::Attachment::Stripper> that we can't work
around.

You cannot modify Flags through the email interface.